改进混合高斯模型在人体跌倒检测中的应用  

Research of human fall detection algorithm based on improved Gaussian mixture mode

作　　者：孙朋 夏飞 张浩 彭道刚[1,2] 马茜 罗志疆 SUN Peng;XIA Fei;ZHANG Hao;PENG Daogang;MA Xi;LUO Zhijiang(College of Automation Engineering, Shanghai University of Electric Power, Shanghai 200090, China;Shanghai Engineering Research Center of Intelligent Management and Control for Power Process, Shanghai 200090, China;CMIS Center, Tongji University, Shanghai 200092, China)

机构地区：[1]上海电力学院自动化工程学院,上海200090 [2]上海发电过程智能管控工程技术研究中心,上海200090 [3]同济大学CMIS中心,上海200092

出　　处：《计算机工程与应用》2017年第20期173-179,共7页Computer Engineering and Applications

基　　金：上海市"科技创新行动计划"高新技术领域科研项目(No.15111106800)

摘　　要：提出了一种基于改进混合高斯模型的分级特征检测算法,对人体跌倒状态进行检测。针对实际目标检测过程中背景更新缓慢,阴影干扰等缺点,通过改进混合高斯模型进行背景更新,并根据阴影区域在HSV颜色空间的特征信息消除阴影干扰。利用人体最小面积外接矩形和垂直外接矩形对检测到的人体目标进行标记,分析目标区域的矩形宽高比、人体质心高度比和人体躯干倾斜角的特征变化。根据各个特征对人体不同状态的判断灵敏度,提出了一种分级特征检测的方法。首先通过人体躯干倾角的特征,判断出人处于非站立状态。接下来依次采用人体质心高度比和矩形宽高比的特征,确认人体处于跌倒状态。实验结果证明,采用提出的方法对人体跌倒状态进行检测,其环境适应性和跌倒检测准确率均高于采用背景差分和直接检测的各种方法。A hierarchical feature detection based on improved Gaussian mixture model is proposed in this paper to detectthe falls.Aiming at the drawbacks of low background updating rate and shadow interference,background model is updatedby improving the Gaussian mixture model and the shadow interference is eliminated through the characteristics of shadowsin the HSV color space.Human minimum area external rectangle and vertical external rectangle are used to detect thehuman body,and the characteristics of rectangle ratio,centroid height ratio and inclination angle are analyzed in the paper.A hierarchical feature detection is proposed based on the different sensitivities of the characteristics.Firstly,the non-uprightpostures are detected by using human inclination angle.Secondly,the falls are confirmed by using the centroid heightratio and rectangle ratio successively.The experimental results show that the fall detection algorithm proposed in thispaper has preferable adaptive performance and higher fall detection accuracy compared with different background subtractionalgorithms and direct detection algorithms.

关 键 词：混合高斯模型 HSV 分级特征 跌倒检测 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]